A fix to address issues with near-surface low relative humidity and sometimes spurious temperature structures (e.g. surface superadiabats of 4K) in thick fog layers, reported by forecasters from OMSZ (Hungary) and the Met Office (UK) will be implemented operationally within Cycle 47r3 with the
06 UTC run on 22 Feb 2022
This fix will go into operations and be available as part of the operational output from this run onward. Users should be aware that this fix also includes an upgrade of the GRIB 2 Master Tables Version from 27 to 28. This change will be transparent in terms of ecCodes usage but might affect those checking the tables version explicitly.
This fix to the model physics addresses these occasional spurious oscillatory low-level q/t structures (low RH and superadiabatic layer) in thick fog conditions in the lowest 150 m of the atmosphere, to ensure realistic well-mixed structures are retained whilst keeping the rest of the model atmosphere untouched. This was technically not a bug, but an occasional instability that was uncovered in Cycle 47r3.
A test of the fix was performed by re-running HRES and ensembles over 15-day periods in summer and winter, which found a neutral impact on near surface scores, and no impact on the rest of the model atmosphere. The events are rare enough that even if they are positively impacted this will not show up as significantly positive in such a short sample. Nonetheless further verification has been made on specific fog cases and results are consistently good.
Evaluation
Hungary Fog 17 Nov 2021
As reported by the Hungarian weather service OMSZ, there was thick fog (lifting in the morning) in Eastern Hungary that was reproduced in the model, albeit with a low near surface relative humidity at model level 137 (10 m) dropping locally below 40% (left plot). With the fix (right plot) the problem is alleviated with a relative humidity of order 90%.
|
|
Examples abridged from the Daily Report of the "Analyst on Duty" at ECMWF:
13 January 2022
It is clear that the fix lowers the temperature in many areas in the southern part of the UK. The biggest difference is at Reading, where the temperature dropped about 4 degrees. The tephigrams with and without the fix show the impact in Reading, where close to the surface the strange drying, and sudden temperature increase with the superadiabat, have disappeared.
|
|
|
|
|
Top row: T+84h HRES 2m temperature forecast, valid for Thu 12z (13 Jan 2022); operational forecast, then with the fix, then the difference between the two.
Bottom row: soundings for Reading, from the operational forecast and then with the fix.
The visibility has changed a little, with a slight reduction of very low visibility values, but overall they are very similar.
To focus on foggy areas, below we show the dew point depression over areas where visibility was less than 100m. It is very clear on the model run comparison plots that the large dew point depression values have disappeared. There still exist some locations with dew point depression up to 3-4 degrees maximum, but these are really very rare and appear only over orographic areas, like the example tephigram example from the Rockies over western Canada, with the largest dew point depression of 4.0 K. However, this does not seem to show any sign of sudden near surface warming, it just happens to be drier on the surface, even with this low visibility.
|
|
|
|
|
Top row: Visibility at T+84h = 12z on Thu 13 Jan 2022 in HRES forecasts from forecast without the fix (operational run) and then with the fix.
Middle row: Dew point depression over areas where the visibility is <100m, same runs as in row above.
Bottom row: a forecast tephigram example for a point in the Rockies in Canada, where the dew point depression is still large (in this example 4 deg C) with <100 m visibility.
15 Jan 2022
On Saturday 15 January, forecasts of fog in HRES were quite good, and even though the affected area was smaller and concentrated in the east of England, fog was observed and it was more widespread than the previous days when fog was over-forecast. The 10m wind from HRES on 13, 14 and 15 January suggests that on the 15th we have clear southwesterly winds which brought in moister air when drier calmer conditions were experienced in the previous days.
|
|
Left: Base time 14 January 00 UTC, valid at T+9h. Visibility in HRES and Observations, and 10 m wind speed from HRES.
Right: Base time 15 January 00 UTC, valid at T+9h. Visibility in HRES and Observations, and 10 m wind speed from HRES, with Visible Satellite image at the valid time. Met Office web site fog warnings are also shown.
The revised model not only compares better to observed visibility but for 15 January also cures the unrealistic dewpoint depression, locally > 10C, in the very foggy area over North Western France.
|
|
|---|---|
|
|
|
|
|
|
|
|
Left column: 14 January 9UTC, Right column: 15 January 9 UTC.
Row 1 - Visibility Observations;
Row 2 - Visibility HRES operational runs;
Row 3 - Visibility HRES runs with fix implemented;
Row 4 - Dewpoint depression HRES operational;
Row 5 - Dewpoint depression with fix implemented.